From: Sebastian Huber <sebastian.huber@embedded-brains.de>
To: newlib@sourceware.org
Subject: [PATCH 03/11] For the case when RB_REMOVE requires a nontrivial
Date: Mon, 19 Oct 2020 18:05:16 +0200 [thread overview]
Message-ID: <20201019160522.15408-4-sebastian.huber@embedded-brains.de> (raw)
In-Reply-To: <20201019160522.15408-1-sebastian.huber@embedded-brains.de>
From: dougm <dougm@FreeBSD.org>
search to find the node to replace the one being removed, restructure to first
remove the replacement node and correct the parent pointers around it, and then
let the all-cases code at the end deal with the parent of the deleted node,
making it point to the replacement node. This removes one or two conditional
branches.
Reviewed by: markj
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D24845
---
newlib/libc/include/sys/tree.h | 52 ++++++++++++++++------------------
1 file changed, 24 insertions(+), 28 deletions(-)
diff --git a/newlib/libc/include/sys/tree.h b/newlib/libc/include/sys/tree.h
index 5c31ff2ea..b1e5e23a7 100644
--- a/newlib/libc/include/sys/tree.h
+++ b/newlib/libc/include/sys/tree.h
@@ -562,48 +562,44 @@ name##_RB_REMOVE_COLOR(struct name *head, struct type *parent, struct type *elm)
attr struct type * \
name##_RB_REMOVE(struct name *head, struct type *elm) \
{ \
- struct type *child, *parent, *old = elm; \
+ struct type *child, *parent, *parent_old, *old = elm; \
int color; \
- if (RB_LEFT(elm, field) == NULL) \
- child = RB_RIGHT(elm, field); \
- else if (RB_RIGHT(elm, field) == NULL) \
- child = RB_LEFT(elm, field); \
- else { \
+ parent_old = parent = RB_PARENT(elm, field); \
+ color = RB_COLOR(elm, field); \
+ if (RB_LEFT(elm, field) == NULL) { \
+ elm = child = RB_RIGHT(elm, field); \
+ if (elm != NULL) \
+ RB_PARENT(elm, field) = parent; \
+ } else if (RB_RIGHT(elm, field) == NULL) { \
+ elm = child = RB_LEFT(elm, field); \
+ RB_PARENT(elm, field) = parent; \
+ } else { \
elm = RB_RIGHT(old, field); \
if ((child = RB_LEFT(elm, field)) == NULL) { \
child = RB_RIGHT(elm, field); \
RB_RIGHT(old, field) = child; \
- RB_PARENT(elm, field) = elm; \
+ parent = elm; \
} else { \
do \
elm = child; \
while ((child = RB_LEFT(elm, field)) != NULL); \
child = RB_RIGHT(elm, field); \
+ parent = RB_PARENT(elm, field); \
+ RB_LEFT(parent, field) = child; \
+ if (child != NULL) \
+ RB_PARENT(child, field) = parent; \
RB_PARENT(RB_RIGHT(old, field), field) = elm; \
} \
RB_PARENT(RB_LEFT(old, field), field) = elm; \
- parent = RB_PARENT(old, field); \
- if (parent != NULL) { \
- if (RB_LEFT(parent, field) == old) \
- RB_LEFT(parent, field) = elm; \
- else \
- RB_RIGHT(parent, field) = elm; \
- } else \
- RB_ROOT(head) = elm; \
+ color = RB_COLOR(elm, field); \
+ elm->field = old->field; \
} \
- parent = RB_PARENT(elm, field); \
- color = RB_COLOR(elm, field); \
- if (child != NULL) \
- RB_PARENT(child, field) = parent; \
- if (parent != NULL) { \
- if (RB_LEFT(parent, field) == elm) \
- RB_LEFT(parent, field) = child; \
- else \
- RB_RIGHT(parent, field) = child; \
- } else \
- RB_ROOT(head) = child; \
- if (elm != old) \
- (elm)->field = (old)->field; \
+ if (parent_old == NULL) \
+ RB_ROOT(head) = elm; \
+ else if (RB_LEFT(parent_old, field) == old) \
+ RB_LEFT(parent_old, field) = elm; \
+ else \
+ RB_RIGHT(parent_old, field) = elm; \
if (color == RB_BLACK) \
name##_RB_REMOVE_COLOR(head, parent, child); \
while (parent != NULL) { \
--
2.26.2
next prev parent reply other threads:[~2020-10-19 16:05 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-19 16:05 [PATCH 00/11] Synchronize <sys/tree.h> with FreeBSD Sebastian Huber
2020-10-19 16:05 ` [PATCH 01/11] Add RB_REINSERT(3), a low overhead alternative to Sebastian Huber
2020-10-19 16:05 ` [PATCH 02/11] Correct the use of RB_AUGMENT in the RB_TREE Sebastian Huber
2020-10-19 16:05 ` Sebastian Huber [this message]
2020-10-19 16:05 ` [PATCH 04/11] RB_REMOVE invokes RB_REMOVE_COLOR either when Sebastian Huber
2020-10-19 16:05 ` [PATCH 05/11] Remove from RB_REMOVE_COLOR some null checks Sebastian Huber
2020-10-19 16:05 ` [PATCH 06/11] To reduce the size of an rb_node, drop the color Sebastian Huber
2020-10-19 16:05 ` [PATCH 07/11] Restore an RB_COLOR macro, for the benefit of Sebastian Huber
2020-10-19 16:05 ` [PATCH 08/11] Fixup r361997 by balancing parens. Duh Sebastian Huber
2020-10-19 16:05 ` [PATCH 09/11] Linuxkpi uses the rb-tree structures Sebastian Huber
2020-10-26 9:52 ` [PATCH 00/11] Synchronize <sys/tree.h> with FreeBSD Corinna Vinschen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201019160522.15408-4-sebastian.huber@embedded-brains.de \
--to=sebastian.huber@embedded-brains.de \
--cc=newlib@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).